multi-agent reinforcement learning (MARL)